High stale ratio on Zilliqa pool when dual-mining ironfish+zil

[mwpastore]

I'm getting a high ratio of stale Zilliqa shares across multiple (small) rigs when dual-mining ironfish+zil.

Here's my rigel 1.4.4 command line on a small rig with two RTX 3060 12GB cards:

$ sudo ./rigel -a ironfish+zil -w mongo \
    -o [1]stratum+ssl://iron.fpmp.net:9999 -u [1]REDACTED \
        --lock-mclock [1]810 --lock-cclock [1]1425 --cclock [1]100 \
    -o [2]zmp://zil.flexpool.io -u [2]REDACTED --zil-countdown \
        --mclock [2]1200 --lock-cclock [2]1350

Here's rigel transitioning into a Zilliqa session on this rig:

[2023-04-26 15:00:33] Zilliqa session started
[2023-04-26 15:00:33] GPU#0: core offset set to 0
[2023-04-26 15:00:33] GPU#1: core offset set to 0
[2023-04-26 15:00:33] GPU#1: memory offset set to 1200
[2023-04-26 15:00:33] GPU#1: locked core set to 1350
[2023-04-26 15:00:33] GPU#0: memory offset set to 1200
[2023-04-26 15:00:33] GPU#0: locked core set to 1350
[2023-04-26 15:00:33] GPU#1: locked memory set to `unlocked`
[2023-04-26 15:00:33] GPU#0: locked memory set to `unlocked`

And here it is mining zil at full bore:

[2023-04-26 15:01:01] +====================== zil(now) =======================+
[2023-04-26 15:01:01] |#|Name    |Hashrate  |Pool      |Acc|Rej|Inv|Eff       |
[2023-04-26 15:01:01] |0|RTX 3060|44.81 MH/s|955.5 KH/s| 43| 18|  0|431.6 KH/W|
[2023-04-26 15:01:01] |1|RTX 3060|44.46 MH/s|800.0 KH/s| 36| 10|  0|432.1 KH/W|
[2023-04-26 15:01:01] +-+--------+----------+----------+---+---+---+----------+
[2023-04-26 15:01:01] |    Total: 89.26 MH/s|1.755 MH/s| 79| 28|  0|431.8 KH/W|
[2023-04-26 15:01:01] +=======================================================+

You can see that the overclock is applied correctly and the hash rate and efficiency are absolutely perfect, but the reject ratio reported by rigel is over 26%. Here's the corresponding stat line on the pool:

I don't believe it's network connectivity or other hardware issue with the rig as I'm getting a similar ratio across several different rigs and Internet uplinks. The ratio of valid Iron Fish shares on this rig in particular is 99.76%, which is more than acceptable.

Any suggestions? Or is it nothing to be concerned about? Thank you in advance!

ETA: I'm testing bzminer now, and if I have the same issue I'll test with a different pool.

[AionIntDevGM]

Yes, Me 2 I have a 3080 on 2 machines same pool, different miners this one has TONS of Stales 28 (13.21%) | 0 (0%) vs Zero on bzminer

[mwpastore]

bzminer does not exhibit this issue using the same pools, overclocks, etc. However, the amount of valid shares over an interval is roughly the same regardless of the mining software used. This leads me to believe that bzminer is filtering out stale shares and not submitting them to the pool whereas rigel is simply sending everything. I'm re-testing rigel with the --cpu-check command-line argument to see if that brings results for the two miners back into alignment.

[AionIntDevGM]

Or Rigel is just triggering alot of stales and BZ is not

[AionIntDevGM]

Used 14.3.1 to test last night and current version of Rigel, I like the interface better on Rigel, but is also seems to be showing more power useage too, I am switching that miner to bz to run for another 24 hrs and compare

[rigelminer]

I believe this bug may be flexpool specific, and since they are going to drop Zil support on May 19 it's probably not worth fixing this bug if it only occurs when mining to flexpool. Are you able to try to switch Zil to another pool and let me know if the issue persists? Thank you.

[rigelminer]

Closing the ticket now as the issue was only present on flexpool which dropped zil support. If it is still present on other pools, please open a new issue.